DNS gaffe leaves spy agency totally under cover
Big website knickers round ankles
The unavailability of the US National Security Agency website on Thursday has been linked to misconfigured DNS (Domain Name System) servers.
Surfers were unable to reach NSA.gov from about 0700 on Thursday because systems used to translate web addresses humans understand to machine-readable IP addresses were playing up, according to an analysis by Arbor Networks.
Danny McPherson, chief research officer with Arbor Networks, explains in detail how the signal intelligence agency's two authoritative DNS servers were temporarily unreachable. The same problem would (potentially) affect MX records and therefore email delivery.
The DNS servers might have been left unreachable for a variety of reasons ranging from router or network misconfiguration to server failure, network outage and response to hostile attack, in decreasing order of probability.
It's unclear what caused the outage, but McPherson's analysis provides evidence that the spy agency made a brace of elementary mistakes with in configuring its DNS systems. For one thing, a web server was run on the same machine (or at least same IP address) as one of the authoritative name server for nsa.gov. Secondly the primary and secondary authoritative name servers are both downstream from the same Qwest edge access router in Washington DC, instead of being properly separated.
YouTube recently experienced availability glitches due to similar problems in the second category, McPherson notes. Failure on YouTube's part to apply industry best practise for running its DNS system left it intermittently difficult to watch the Star Wars kid, and similar video gems. The temporary nonavailability of the NSA website is a whole lot more serious.
And since the agency is charged with advising US firms how to safeguard their internet infrastructures, as well as actively attempting to break into the systems of intelligence targets, it's only right that the NSA is held to far higher standards than YouTube.
NSA techies have now restored the site, but everyone involved in the outage should be thoroughly ashamed of themselves. ®
Sponsored: DevOps and continuous delivery